Environmentalist

Yes, and this really depends. A 1980 car, well stored and taken care when not in use, which is used only once a fortnight for getting groceries in the village 5km away, probably is in good technical state. If it were not for the political will ("reduce pollution") that gets old cars banned, it would probably be more ecological as economical to continue using that car for some years more, instead of trashing a perfectly good item for a newly made one.
You miss the point. It is all about creating consumption!

We are at a point where most people have everything they need. The biggest issue for our economy is to keep up the necessary ever-increasing growth of consumption. So, the environment story is here just to make people throw away more stuff and buy new one.

This is a "double feature" scheme: on one hand the people are told that they must do something for the environment (and obviousely that is a moral imperative, so you cannot say anything against it), but then, if you look closer, those who are in power are all associated with investors, they make up these moral imperatives, while in fact they only have to provide for the profits, for "creating new markets".

Donald Trump called them the "establishment", I call it "feudal-socialism": it works like socialism when treating the people as government-owned cattle: we are unfree, we are told what to think, we are told what to do, all with moral imperatives. And it works like feudalism when securing the profits for the elite. Just like in the middle ages, when religious morals were employed by the gentry to force the people to comply with what was said to be god's will. Now it's no longer god's will, but it's still the same scheme of unquestionable morals abused to subdue the people.

The essential point is that it is the same people. You can complain about how bad the world is, and how endangered the environment, and how troublesome our over-consumption, and how poor the people in Africa, and so on and so on, and you will be perfectly welcome with all that sermon. But what you're not allowed to say is that it's the same people who actively create those problems and who tell us that we must feel guilty because of them!

The priest who talks about how sinful the world is and that we all must repent, is the same person who actively rapes little boys in the sacristy. The aid agencies who tell us that we must care for all the refugees are the same persons who actively send midwifes (instead of prevention consultants) to Africa to create more over-population. The people who complain about the climate catastrophy are the same people who make big money with carbondioxide certificates. And so on and so on. But that's what you must not say - because if you say that openly, and attach the proper names, then -guess what- you are a nazi.
 
You don't get them anymore. Given I get my intended haswell-board, it will not have any PCI/PCI-x slots, so I cannot use my fxp 100Mbit adapters anymore.
You don’t have to use 100 Mbps adapters. You can run 1 Gbps adapters at 100 Mbps, too, and this will also save power.
ifconfig igb0 media 100baseTX mediaopt full-duplex

Some NICs support power saving modes, for example see the tunable hw.em.smart_pwr_down in em(4). Also, some NICs will power down a port automatically when the interface is marked “down” (sometimes you have to disable WOL for this to work).

Another thing worth mentioning is the sysctl hw.pci.do_power_nodriver. It causes power to be switched off on PCI connections that have no driver loaded.
 
I remember people telling me to get rid of my old TDI and buy a shiny new car, "'For the environment'. When driven carefully, I got it down to 3.7l/100km. Normally, 4.5 and when I was in a hurry still <9.
Producing a new one would consume enough energy to drive it a lot more than 150.000km. And before you ask, it had all the trimmings. The greenest thing to do was drive it as long as possible. Had to sell it at 300.000km. :(

So, it should be best to use old HW for as long as possible. The energy cost at your home should not be considered but the whole 'ecological backpack'.
Yes, but it’s very difficult to estimate that backpack.

For example, an older PC of mine took one to two days for transcoding a movie (I do this a lot), and consumed 280 W during all that time. My new PC takes about an hour for the same thing and consumes 120 W under load. It’s roughly a factor of 80 (!) in total power consumption per movie, i.e. 10 kWh vs. 120 Wh. This is quite a difference. But whether this makes up for the “ecological backpack”, I have no idea. But even if it didn’t, I prefer the newer machine for the fact that it makes much less noise 90 % of the time.

For me personally, quality of life is an important factor, too.
 
This is all in all a komplex topic.
We did this better better better race for some time. As a result, we have made progress like never before. Compare the life of you to your grandparents. And rhen their grandparents. The unemployed today can live better than the kings of the past. Ol' man Rockefeller could not talk to all the wold like we can. Queen Victoria did not get fresh produce in winter, not to mention a fridge.
That there is an even bigger gap between the unemployed today ad f.e. Elon - that is a price we need to py for this.
What it does with the planet is part of the price, and we might very well find us in a place where we cannot afford that any longer. IMHO we passed that point already. We need to drive technology ahead now, because nothing else has a potential to save us. And whoever doubts that the price is too hight when the bill comes around, I have some seafront estate in Florida you might find interesting.
 
You don’t have to use 100 Mbps adapters. You can run 1 Gbps adapters at 100 Mbps, too, and this will also save power.
ifconfig igb0 media 100baseTX mediaopt full-duplex

Some NICs support power saving modes, for example see the tunable hw.em.smart_pwr_down in em(4). Also, some NICs will power down a port automatically when the interface is marked “down” (sometimes you have to disable WOL for this to work).

Another thing worth mentioning is the sysctl hw.pci.do_power_nodriver. It causes power to be switched off on PCI connections that have no driver loaded.
Yes, been thru that, didn't find much that would make the device (it was some elderly intel quad-port 1G) become not so hot (I was rather bothered with possible airflow planning than with energy consumption).

What seems more intereting is this (and most pci devices have that):
Code:
fxp3@pci0:5:7:0:        class=0x020000 card=0xb1640e11 chip=0x12298086 rev=0x08 hdr=0x00
    cap 01[dc] = powerspec 2  supports D0 D1 D2 D3  current D0

How do we switch these D-states?
 
This is all in all a komplex topic.
It is complex. It is futurology at its best, and it should be done a lot more, and with a much wider horizon.
We did this better better better race for some time. As a result, we have made progress like never before. Compare the life of you to your grandparents. And rhen their grandparents. The unemployed today can live better than the kings of the past. Ol' man Rockefeller could not talk to all the wold like we can. Queen Victoria did not get fresh produce in winter, not to mention a fridge.
That there is an even bigger gap between the unemployed today ad f.e. Elon - that is a price we need to py for this.
Thats a problem for the socialists - they always complain as long as not all people are even. I see that only as a problem of greed and envy, i.e. no problem at all: let Elon and folks have as much as they want, and let them be happy with it - as long as I do not need to starve and freeze, thats all fine with me. The problem only starts when these guys start to take away our freedom, our freedom to live our lifes as we like it, and make us behave as puppies to their schemes.

But there is something else (it was already somewhere in Frank Herbert's Dune, in that huge amount of highly valuable side-notes and mentions): in a highly developed economy you will get more and more fancy stuff, and you will have ever greater difficulty to get some of the basic necessary stuff.
Consider you need just a piece of metal, to fix something. Where do you get that? Usually nowhere - instead you harvest it from some scrap, because it is too simple to be on sale anywhere.
Other example: the HP5p printer. It's perfect, it works forever. And you don't get such a thing anymore. You get a "newer and more modern" printer every year (and they all do just the same: print), and you will throw them away after a few years because supplement are no longer available - but you don't get a basic solid printer that can work for some 20 years, because that is no longer on sale anywhere.
More troublesome issue: medical support. There is more and more highly delicate crap, but no longer means to properly treat people in a basic and solid fashion. Ivan Illich (research on that guy, it's valuable) said we actually need only about 25 different medications to get along. The remainder is a vastly inflated market for the only purpose to make money. We are actually laboratory rats abused by the healthcare industry for all kinds of treatments, we are kept in dependency and told that this would be improvement.

What it does with the planet is part of the price, and we might very well find us in a place where we cannot afford that any longer. IMHO we passed that point already. We need to drive technology ahead now, because nothing else has a potential to save us. And whoever doubts that the price is too hight when the bill comes around, I have some seafront estate in Florida you might find interesting.
I don't think the actual problem is that we are in a conflict with the planet - that is, with nature. Because, we ARE nature. I rather think the actual problem is that this advance we are undergoing is very difficult to stabilize in some healthy fashion. We are in a conflict not with nature or the planet, but with our own advance - it is a runaway process.
But then, this is nothing new at all: any highly developed civilisations have entered a state where they would fall apart due to inner corruption. And this again IS nature - the ancient vedic philosophers knew that and termed it Kali Yuga. And yes, this will destroy our civilisation, but this is nothing bad, it has happened before and will happen again anyway.
 
We once housed visitors from japan. A highly advanced nation. What did they buy for gifts to bring back home? Quality screwdrivers, swiss army knives, multitools, field glasses. Sandals. They emtied the tool shops for some cities around. Because that stuff was not available for sale back home.
 
Yes, been thru that, didn't find much that would make the device (it was some elderly intel quad-port 1G) become not so hot (I was rather bothered with possible airflow planning than with energy consumption).

What seems more intereting is this (and most pci devices have that):
Code:
fxp3@pci0:5:7:0:        class=0x020000 card=0xb1640e11 chip=0x12298086 rev=0x08 hdr=0x00
    cap 01[dc] = powerspec 2  supports D0 D1 D2 D3  current D0

How do we switch these D-states?
You can suspend unused devices with devctl(8).
 
You can suspend unused devices with devctl(8).
Wow, this works!
One can even suspend used devices, it's like unplugging the wire, they keep the routing when coming back. :)

Code:
# devctl suspend alc0
# pciconf -lc alc0
alc0@pci0:3:0:0:        class=0x020000 card=0x85071043 chip=0x10911969 rev=0x10 hdr=0x00
    cap 01[40] = powerspec 3  supports D0 D3  current D3
    cap 10[58] = PCI-Express 1 endpoint max data 128(4096)
                 link x1(x1) speed 2.5(2.5) ASPM disabled(L0s/L1) ClockPM disabled
    cap 05[c0] = MSI supports 16 messages, 64 bit, vector masks
    cap 11[d8] = MSI-X supports 16 messages, enabled
                 Table in map 0x10[0x2000], PBA in map 0x10[0x3000]
    ecap 0001[100] = AER 1 0 fatal 1 non-fatal 0 corrected
    ecap 0003[180] = Serial 1 ffda320908606eff

Actually it seems to work nice for everything except the card where I would really like to use it. :/
This one works:
igb1@pci0:1:0:1: class=0x020000 card=0xa04c8086 chip=0x10c98086 rev=0x01 hdr=0x00
cap 01[40] = powerspec 3 supports D0 D3 current D3
cap 05[50] = MSI supports 1 message, 64 bit, vector masks
cap 11[70] = MSI-X supports 10 messages, enabled
Table in map 0x1c[0x0], PBA in map 0x1c[0x2000]
cap 10[a0] = PCI-Express 2 endpoint max data 256(512) FLR NS
link x4(x4) speed 2.5(2.5) ASPM disabled(L0s/L1)
ecap 0001[100] = AER 1 0 fatal 0 non-fatal 1 corrected
ecap 0003[140] = Serial 1 90e2baffff7e80ac
ecap 000e[150] = ARI 1
ecap 0010[160] = SR-IOV 1 IOV disabled, Memory Space disabled, ARI disabled
0 VFs configured out of 8 supported
First VF RID Offset 0x0180, VF RID Stride 0x0002
VF Device ID 0x10ca
Page Sizes: 4096 (enabled), 8192, 65536, 262144, 1048576, 4194304


While this doesn't - it configures suspend, but doesn't change the D0 mode (and neither the heat dissipation):
em3@pci0:4:0:1: class=0x020000 card=0x11bc108e chip=0x10bc8086 rev=0x06 hdr=0x00
cap 01[c8] = powerspec 2 supports D0 D3 current D0
cap 05[d0] = MSI supports 1 message, 64 bit enabled with 1 message
cap 10[e0] = PCI-Express 1 endpoint max data 256(256) NS
link x4(x4) speed 2.5(2.5) ASPM disabled(L0s)
ecap 0001[100] = AER 1 0 fatal 1 non-fatal 0 corrected
ecap 0003[140] = Serial 1 001517ffffdc84ca
 
There is an ongoing discussion about programming languages.

One question I never found being discussed is the energy-efficiency of programming languages.
But in terms of total cost of ownership, this might be(come) a relevant factor.
For example, I liked much how drhowarddrfine told about his company, specialized on writing web server applications in C, mainly for speed.
What if one company has 1000 servers mainly working on PHP or other interpreting stuff? How many would be needed if the applications were written in, say, C? How big would the energy savings be financially?

I sometimes think about how much energy consumption would be needed to compensate for more man-hour-cost of developing energy-efficient software.
For rarely-used software that doesn't matter much, but for big heavy-duty services it might be relevant.

Does anybody know of research in this direction?
 
IIRC the computer scientists have a rule of thumb: a good algorithm beats a good implementation.
Traditionally, SW engineers treat runtime=cost. FMLU of tuning, it went like this:
  1. measure the execution times in your scripted, interpreted application for your average workload.
  2. Identify the most costly modules/routines.
  3. For these: research if there's a better algorithm to do that. If yes, implement & go back to 1.
  4. Have a close look on the data structures used & optimize them. Goto 1.
  5. If it's still too slow, rewrite in the most suitable compiled language. Whatever "most suitable" means, depends... Often the choice was C/C++.
So you also want to measure energy consumption instead. Nice idea.
Not only PHP is ugly, it's broken by design because it mixes application data & logic, that's a no-go; using it is a criminal act against proven & reliable principles of SW engineering. Thus the major benefit to replace that crap with a correctly designed webapp framework will be much greater than only in terms of energy.
Nowadays we have JIT compilers, which can adopt their optimization to runtime parameters, thus hand-crafted C/C++ modules are oldschool, usually inferior to JIT compilation & thus not needed, except in rare special cases. Say what you want against JavaScript, AFAIK these node.js wizzards wrote a damn good piece of SW.
 
Nowadays we have JIT compilers, which can adopt their optimization to runtime parameters, thus hand-crafted C/C++ modules are oldschool, usually inferior to JIT compilation & thus not needed, except in rare special cases. Say what you want against JavaScript, AFAIK these node.js wizzards wrote a damn good piece of SW.
Yes, but what I wonder is how efficient is it to load and run the JIT compilers gazillions of times.
No matter how efficient, this repetitive work with always the same result sums up both in energy and cost.
 
But in terms of total cost of ownership, this might be(come) a relevant factor.
...
Does anybody know of research in this direction?
Lots. In the 90s and 2000s, I saw lots of work on this for the mobile (cell phone) applications, and there were lots of publications about that. I don't know whether that research is still ongoing, I have lost interest in it. And for server applications: there are lots of companies that have millions of servers. There a small improvement (like making a compiler's generated code 1% faster) have large effects. And the big companies that have thousands or tens of thousands of software engineers also measure programmer productivity. So the tradeoff between ease-of-development and efficient execution is carefully measured and managed.
 
Back
Top